From f8dc3abe36615ed71eec42f39ab1b4a31e5d31f3 Mon Sep 17 00:00:00 2001
From: cheatfate <eugene.kabanov@status.im>
Date: Wed, 20 Mar 2019 11:41:37 +0200
Subject: [PATCH] Add MultiAddress pattern matching procedures
 (go-multiaddr-fmt) with tests. Add some comments.

---
 libp2p/ipnet/iface.nim     |   3 +
 libp2p/ipnet/ipnet.nim     |   1 +
 libp2p/multiaddress.nim    | 120 ++++++++++++++++++++++++++++++++++++-
 tests/testmultiaddress.nim |  97 ++++++++++++++++++++++++++++++
 4 files changed, 220 insertions(+), 1 deletion(-)

diff --git a/libp2p/ipnet/iface.nim b/libp2p/ipnet/iface.nim
index 0eb328a4a..d1b831ce4 100644
--- a/libp2p/ipnet/iface.nim
+++ b/libp2p/ipnet/iface.nim
@@ -6,6 +6,9 @@
 ## at your option.
 ## This file may not be copied, modified, or distributed except according to
 ## those terms.
+
+## This module implements cross-platform network interfaces list.
+## Currently supported OSes are Windows, Linux, MacOS, BSD(not tested).
 import algorithm
 from strutils import toHex
 import ipnet
diff --git a/libp2p/ipnet/ipnet.nim b/libp2p/ipnet/ipnet.nim
index cd781db72..2a1206822 100644
--- a/libp2p/ipnet/ipnet.nim
+++ b/libp2p/ipnet/ipnet.nim
@@ -7,6 +7,7 @@
 ## This file may not be copied, modified, or distributed except according to
 ## those terms.
 
+## This module implements various IP network utility procedures.
 import endians, strutils
 import chronos
 export chronos
diff --git a/libp2p/multiaddress.nim b/libp2p/multiaddress.nim
index 3132b74b7..96e9211ca 100644
--- a/libp2p/multiaddress.nim
+++ b/libp2p/multiaddress.nim
@@ -10,7 +10,7 @@
 ## This module implements MultiAddress.
 import tables, strutils, net
 import multicodec, multihash, multibase, transcoder, base58, base32, vbuffer
-import peer
+from peer import PeerID
 
 {.deadCodeElim:on.}
 
@@ -27,6 +27,18 @@ type
   MultiAddress* = object
     data*: VBuffer
 
+  MaPatternOp* = enum
+    Eq, Or, And
+
+  MaPattern* = object
+    operator*: MaPatternOp
+    args*: seq[MaPattern]
+    value*: MultiCodec
+
+  MaPatResult* = object
+    flag*: bool
+    rem*: seq[MultiCodec]
+
   MultiAddressError* = object of Exception
 
 proc ip4StB(s: string, vb: var VBuffer): bool =
@@ -219,6 +231,21 @@ proc dnsVB(vb: var VBuffer): bool =
     if s.find('/') == -1:
       result = true
 
+proc pEq(codec: string): MaPattern =
+  ## ``Equal`` operator for pattern
+  result.operator = Eq
+  result.value = multiCodec(codec)
+
+proc pOr(args: varargs[MaPattern]): MaPattern =
+  ## ``Or`` operator for pattern
+  result.operator = Or
+  result.args = @args
+
+proc pAnd(args: varargs[MaPattern]): MaPattern =
+  ## ``And`` operator for pattern
+  result.operator = And
+  result.args = @args
+
 const
   TranscoderIP4* = Transcoder(
     stringToBuffer: ip4StB,
@@ -349,6 +376,40 @@ const
     )
   ]
 
+  DNS4* = pEq("dns4")
+  DNS6* = pEq("dns6")
+  IP4* = pEq("ip4")
+  IP6* = pEq("ip6")
+  DNS* = pOr(pEq("dnsaddr"), DNS4, DNS6)
+  IP* = pOr(IP4, IP6)
+  TCP* = pOr(pAnd(DNS, pEq("tcp")), pAnd(IP, pEq("tcp")))
+  UDP* = pOr(pAnd(DNS, pEq("udp")), pAnd(IP, pEq("udp")))
+  UTP* = pAnd(UDP, pEq("utp"))
+  QUIC* = pAnd(UDP, pEq("quic"))
+
+  Unreliable* = pOr(UDP)
+
+  Reliable* = pOr(TCP, UTP, QUIC)
+
+  IPFS* = pAnd(Reliable, pEq("p2p"))
+
+  HTTP* = pOr(
+    pAnd(TCP, pEq("http")),
+    pAnd(IP, pEq("http")),
+    pAnd(DNS, pEq("http"))
+  )
+
+  HTTPS* = pOr(
+    pAnd(TCP, pEq("https")),
+    pAnd(IP, pEq("https")),
+    pAnd(DNS, pEq("https"))
+  )
+
+  WebRTCDirect* = pOr(
+    pAnd(HTTP, pEq("p2p-webrtc-direct")),
+    pAnd(HTTPS, pEq("p2p-webrtc-direct"))
+  )
+
 proc initMultiAddressCodeTable(): Table[MultiCodec,
                                         MAProtocol] {.compileTime.} =
   result = initTable[MultiCodec, MAProtocol]()
@@ -527,6 +588,12 @@ proc `$`*(value: MultiAddress): string =
   if len(parts) > 0:
     result = "/" & parts.join("/")
 
+proc protocols*(value: MultiAddress): seq[MultiCodec] =
+  ## Returns list of protocol codecs inside of MultiAddress ``value``.
+  result = newSeq[MultiCodec]()
+  for item in value.items():
+    result.add(item.protoCode())
+
 proc hex*(value: MultiAddress): string =
   ## Return hexadecimal string representation of MultiAddress ``value``.
   result = $(value.data)
@@ -715,3 +782,54 @@ proc isWire*(ma: MultiAddress): bool =
         break
   except:
     result = false
+
+proc matchPart(pat: MaPattern, protos: seq[MultiCodec]): MaPatResult =
+  var empty: seq[MultiCodec]
+  var pcs = protos
+  if pat.operator == Or:
+    for a in pat.args:
+      let res = a.matchPart(pcs)
+      if res.flag:
+        return MaPatResult(flag: true, rem: res.rem)
+    result = MaPatResult(flag: false, rem: empty)
+  elif pat.operator == And:
+    if len(pcs) < len(pat.args):
+      return MaPatResult(flag: false, rem: empty)
+    for i in 0..<len(pat.args):
+      let res = pat.args[i].matchPart(pcs)
+      if not res.flag:
+        return MaPatResult(flag: false, rem: res.rem)
+      pcs = res.rem
+    result = MaPatResult(flag: true, rem: pcs)
+  elif pat.operator == Eq:
+    if len(pcs) == 0:
+      return MaPatResult(flag: false, rem: empty)
+    if pcs[0] == pat.value:
+      return MaPatResult(flag: true, rem: pcs[1..^1])
+    result = MaPatResult(flag: false, rem: empty)
+
+proc match*(pat: MaPattern, address: MultiAddress): bool =
+  ## Match full ``address`` using pattern ``pat`` and return ``true`` if
+  ## ``address`` satisfies pattern.
+  var protos = address.protocols()
+  let res = matchPart(pat, protos)
+  result = res.flag and (len(res.rem) == 0)
+
+proc matchPartial*(pat: MaPattern, address: MultiAddress): bool =
+  ## Match prefix part of ``address`` using pattern ``pat`` and return
+  ## ``true`` if ``address`` starts with pattern.
+  var protos = address.protocols()
+  let res = matchPart(pat, protos)
+  result = res.flag
+
+proc `$`*(pat: MaPattern): string =
+  ## Return pattern ``pat`` as string.
+  var sub = newSeq[string]()
+  for a in pat.args:
+    sub.add($a)
+  if pat.operator == And:
+    result = sub.join("/")
+  elif pat.operator == Or:
+    result = "(" & sub.join("|") & ")"
+  elif pat.operator == Eq:
+    result = $pat.value
diff --git a/tests/testmultiaddress.nim b/tests/testmultiaddress.nim
index c26a781bb..b75676720 100644
--- a/tests/testmultiaddress.nim
+++ b/tests/testmultiaddress.nim
@@ -1,6 +1,12 @@
 import unittest
 import ../libp2p/multiaddress
 
+type
+  PatternVector = object
+    pattern: MaPattern
+    good: seq[string]
+    bad: seq[string]
+
 const
   SuccessVectors = [
     "/ip4/1.2.3.4",
@@ -203,6 +209,88 @@ const
     "/ip4/127.0.0.1/udp/1234/quic"
   ]
 
+  PatternVectors = [
+    PatternVector(pattern: IP,
+      good: @["/ip4/0.0.0.0", "/ip6/fc00::"],
+      bad: @["/ip4/0.0.0.0/tcp/555", "/udp/789/ip6/fc00::"]
+    ),
+    PatternVector(pattern: TCP,
+      good: @["/ip4/0.0.7.6/tcp/1234", "/ip6/::/tcp/0"],
+      bad: @["/tcp/12345", "/ip6/fc00::/udp/5523/tcp/9543"]
+    ),
+    PatternVector(pattern: UDP,
+      good: @["/ip4/0.0.7.6/udp/1234", "/ip6/::/udp/0"],
+      bad: @["/udp/12345", "/ip6/fc00::/tcp/5523/udp/9543"],
+    ),
+    PatternVector(pattern: UTP,
+      good: @["/ip4/1.2.3.4/udp/3456/utp", "/ip6/::/udp/0/utp"],
+      bad: @[
+        "/ip4/0.0.0.0/tcp/12345/utp",
+        "/ip6/fc00::/ip4/0.0.0.0/udp/1234/utp",
+        "/utp"
+      ]
+    ),
+    PatternVector(pattern: QUIC,
+      good: @["/ip4/1.2.3.4/udp/1234/quic", "/ip6/::/udp/1234/quic"],
+      bad: @[
+        "/ip4/0.0.0.0/tcp/12345/quic",
+        "/ip6/fc00::/ip4/0.0.0.0/udp/1234/quic",
+        "/quic"
+      ]
+    ),
+    PatternVector(pattern: IPFS,
+      good: @[
+        "/ip4/1.2.3.4/tcp/1234/ipfs/QmaCpDMGvV2BGHeYERUEnRQAwe3N8SzbUtfsmvsqQLuvuJ",
+        "/ip6/::/tcp/1234/ipfs/QmaCpDMGvV2BGHeYERUEnRQAwe3N8SzbUtfsmvsqQLuvuJ",
+        "/ip6/::/udp/1234/utp/ipfs/QmaCpDMGvV2BGHeYERUEnRQAwe3N8SzbUtfsmvsqQLuvuJ",
+        "/ip4/0.0.0.0/udp/1234/utp/ipfs/QmaCpDMGvV2BGHeYERUEnRQAwe3N8SzbUtfsmvsqQLuvuJ",
+      ],
+      bad: @[
+        "/ip4/1.2.3.4/ipfs/QmaCpDMGvV2BGHeYERUEnRQAwe3N8SzbUtfsmvsqQLuvuJ",
+        "/ip6/::/ipfs/QmaCpDMGvV2BGHeYERUEnRQAwe3N8SzbUtfsmvsqQLuvuJ",
+        "/tcp/123/ipfs/QmaCpDMGvV2BGHeYERUEnRQAwe3N8SzbUtfsmvsqQLuvuJ",
+        "/ip6/::/udp/1234/ipfs/QmaCpDMGvV2BGHeYERUEnRQAwe3N8SzbUtfsmvsqQLuvuJ",
+        "/ip6/::/utp/ipfs/QmaCpDMGvV2BGHeYERUEnRQAwe3N8SzbUtfsmvsqQLuvuJ",
+        "/ipfs/QmaCpDMGvV2BGHeYERUEnRQAwe3N8SzbUtfsmvsqQLuvuJ",
+      ]
+    ),
+    PatternVector(pattern: DNS,
+      good: @["/dnsaddr/example.io", "/dns4/example.io", "/dns6/example.io"],
+      bad: @["/ip4/127.0.0.1"],
+    ),
+    PatternVector(pattern: WebRTCDirect,
+      good: @[
+        "/ip4/1.2.3.4/tcp/3456/http/p2p-webrtc-direct",
+        "/ip6/::/tcp/0/http/p2p-webrtc-direct"
+      ],
+      bad: @[
+        "/ip4/0.0.0.0", "/ip6/fc00::", "/udp/12345",
+        "/ip6/fc00::/tcp/5523/udp/9543"
+      ]
+    ),
+    PatternVector(pattern: HTTP,
+      good: @[
+        "/ip4/1.2.3.4/http", "/dns4/example.io/http",
+        "/dns6/::/tcp/7011/http", "/dnsaddr/example.io/http",
+        "/ip6/fc00::/http"
+      ],
+      bad: @[
+        "/ip4/1.2.3.4/https", "/ip4/0.0.0.0/tcp/12345/quic",
+        "/ip6/fc00::/tcp/5523"
+      ]
+    ),
+    PatternVector(pattern: HTTPS,
+      good: @[
+        "/ip4/1.2.3.4/https", "/dns4/example.io/https",
+        "/dns6/::/tcp/7011/https", "/ip6/fc00::/https"
+      ],
+      bad: @[
+        "/ip4/1.2.3.4/http", "/ip4/0.0.0.0/tcp/12345/quic",
+        "/ip6/fc00::/tcp/5523"
+      ]
+    )
+  ]
+
 suite "MultiAddress test suite":
 
   test "go-multiaddr success test vectors":
@@ -264,3 +352,12 @@ suite "MultiAddress test suite":
       check:
         hex(a) == PathExpects[i]
         $a == PathVectors[i]
+
+  test "MultiAddress pattern matching test vectors":
+    for item in PatternVectors:
+      for gitem in item.good:
+        var a = MultiAddress.init(gitem)
+        check item.pattern.match(a) == true
+      for bitem in item.bad:
+        var a = MultiAddress.init(bitem)
+        check item.pattern.match(a) == false